Editing Module:Engvar

-- This module implements Template:Engvar.
-- Template:Engvar is to be build into the template (like an infobox), with default & variant spellings defined.
-- That template should also allow parameter '|engvar=' for the editor (article page).
-- The module/template Engvar then returns the spelling variant as is set in that article (for example '|engvar=en-GB').
-- The defaultWord is returned, unless the engvar input hits on a defined (en-XX) variant word.
local p = {}
local getArgs = require('Module:Arguments').getArgs
local yesno = require('Module:Yesno')
local gsub = string.gsub
local lower = string.lower
local upper = string.upper

function p.variants(frame)
	local args = getArgs(frame)
	return p._variants(args)
end

local function fallback(args, useLang)
	local cwealthLangs = { ['en-OED'] = 1, ['en-IUPAC'] = 1,
		['en-AU'] = 1, ['en-CA'] = 1, ['en-EI'] = 1, ['en-HK'] = 1,
		['en-IN'] = 1, ['en-JM'] = 1, ['en-NZ'] = 1, ['en-PK'] = 1,
		['en-SCO'] = 1,	['en-SG'] = 1, ['en-ZA'] = 1 }
	if (useLang == 'en-IUPAC' or useLang == 'en-OED') and (args['en-OED'] or args['en-CA']) then
		return args['en-OED'] or args['en-CA']
	elseif useLang == 'en-CA' and (args['en-OED'] or args['en-US']) then
		return args['en-OED'] or args['en-US']
	elseif cwealthLangs[useLang] and args['en-GB'] then
		return args['en-GB']
	end
	
	return args.defaultWord or ''
end

function p._variants(args)
	local returnWord = nil
	local auto = false
	
	local defaultWord = args.defaultWord or ''
	local engvar = args.engvar

	if (lower(engvar or '') == 'auto') then
		engvar = require('Module:Engvar/detect').detect()
		auto = true
	end

	if engvar == nil then
		-- Nothing to look for; use defaultWord right away
		returnWord = defaultWord
	elseif args.defaultLang == gsub(lower(engvar), '^en%-(%w%w)$', formatISO) then
		-- By the defaultLang, the defaultWord is asked
		returnWord = defaultWord
	else
		returnWord = args[gsub(lower(engvar), '^en%-(%w%w)$', formatISO)]
	end

	if returnWord == nil then
		-- No hit so far. Search by words in the engvar entered, checking the list
		local useLang = engvarLang(engvar)
		if useLang == nil then
			returnWord = defaultWord .. addMaintCat(args)
		elseif args[useLang] then
			returnWord = args[useLang]
		elseif yesno(args.fallback or auto, auto) then --default to fallback if auto, otherwise default to no fallback
			returnWord = fallback(args, useLang)
		else
			returnWord = defaultWord
		end
	end

	return returnWord
end

-- Returns arguments and intermediate result. Plus the template result, in front.
function p.explain(frame)
	local args = getArgs(frame)
	local ret = {}
	local engvar = args.engvar or ''
	table.insert(ret, '\n\n: Settings:')
	table.insert(ret, 'defaultWord=' .. (args.defaultWord or '') .. '; ')
	table.insert(ret, 'defaultLang=' .. (args.defaultLang or '') .. '; ')
	table.insert(ret, 'engvarCat=' .. (args.engvarCat or '[default:yes]') .. '; ')
	table.insert(ret, 'engvarCatSort=' .. (args.engvarCatSort or ''))
	table.insert(ret, 'fallback=' .. (args.fallback or ''))
	table.insert(ret, '\n\n: engvar=' .. engvar .. ' [input] ')
	if (lower(engvar) == 'auto') then
		engvar = require('Module:Engvar/detect').detect() or ''
		table.insert(ret, 'engvar=' .. engvar .. ' [detected] ')
	end
	local useLang = engvarLang(engvar)
	table.insert(ret, ' => Engvar code [used]: >' .. (useLang or '') .. '<.')
	
	for k, v in pairs(args) do
		if k == 'en-UK' then
			table.insert(ret, k .. ' ? better: use "en-GB"; ')
		elseif k == 'en-SA' then
			table.insert(ret, k .. ' ? misleading; use "en-ZA"; ')
		end
		
		if k == 'defaultWord' then
		elseif k == 'defaultLang' then
		elseif k == 'engvar' then
		elseif k == 'engvarcat' then
		elseif k == gsub(lower(k), '^en%-(%w%w)$', formatISO) then
			table.insert(ret, k .. '=' .. v .. '; ')
		else
			table.insert(ret, k .. ' [not standard:]=' .. v .. '; ')
		end

	end
	return engvar .. ' => ' .. p._variants(args) .. table.concat(ret, ' ')
end

-- Turn a match into pattern 'en-XX'
function formatISO(country)
	return ('en-' .. upper(country) or '')
end

function engvarLang(searchEngvar)
-- Search verbose language identifiers to ISO-format 'en-XX'
-- Assumed: not a blank string '' to search
	local match = string.match
	local langs = {
		us = 'US', usa = 'US',
		au = 'AU', aus = 'AU',
		gb = 'GB', gbr = 'GB', uk = 'GB',
		ca = 'CA', can = 'CA',
		ei = 'EI', irl = 'EI',
		hk = 'HK', hkg = 'HK',
		['in'] = 'IN', ind = 'IN', --'in' is a reserved keyword and needs to be in quotes
		jm = 'JM', jam = 'JM',
		mw = 'MW', mwi = 'MW',
		nz = 'NZ', nzl = 'NZ',
		ng = 'NG', nga = 'NG',
		pk = 'PK', pak = 'PK',
		ph = 'PH', phl = 'PH',
		sco = 'SCO',
		sg = 'SG', sgp = 'SG',
		za = 'ZA', zaf = 'ZA',
		tt = 'TT', tto = 'TT',
		oed = 'OED'};
	searchEngvar = gsub(searchEngvar, '^%s*en%-(.*)', '%1') -- rm any opening 'en-'.
	searchEngvar = gsub(lower(searchEngvar), '[%s%(%)%-]', '') -- To lc, remove all: (, ) , ws, hyphen.
	--normalize language codes so we don't trigger the maintenence category on valid codes
	local useLang = langs[searchEngvar]
	if useLang then
		return 'en-'..useLang
	--Special codes
	elseif match(searchEngvar, 'oxford') then
			useLang = 'OED' -- 'oxford' to catch before anyting 'british'
	elseif match(searchEngvar, 'iupac') then
			useLang = 'IUPAC' -- chemistry
	-- Very often used
	elseif match(searchEngvar, 'british') then
			useLang = 'GB'
	elseif match(searchEngvar, 'unitedstates')
		or match(searchEngvar, 'american') then
			useLang = 'US'
	elseif match(searchEngvar, 'australia') then
			useLang = 'AU'
	-- Often used
	elseif match(searchEngvar, 'india') then
			useLang = 'IN'
	elseif match(searchEngvar, 'newzealand') then
			useLang = 'NZ'
	elseif match(searchEngvar, 'southafrica') then -- not: SA
			useLang = 'ZA'
	elseif  match(searchEngvar, 'canad') then --'canada' or 'canadian'
			useLang = 'CA'
	elseif match(searchEngvar, 'hiberno')
		or match(searchEngvar, 'ireland')
		or match(searchEngvar, 'irish') then
			useLang = 'EI'
	elseif match(searchEngvar, 'hongkong') then
			useLang = 'HK'
	-- Less often used
	elseif match(searchEngvar, 'jamaica') then
			useLang = 'JM'
	elseif match(searchEngvar, 'malawi') then
			useLang = 'MW'
	elseif match(searchEngvar, 'nigeria') then
			useLang = 'NG'
	elseif match(searchEngvar, 'pakistan') then
			useLang = 'PK'
	elseif match(searchEngvar, 'philippine') then
			useLang = 'PH'
	elseif match(searchEngvar, 'scotland')
		or match(searchEngvar, 'scottish') then
			useLang = 'SCO' -- Has no alpha-2 code; not 'scotch'
	elseif match(searchEngvar, 'singapore') then
			useLang = 'SG'
	elseif match(searchEngvar, 'trinidad')
		or match(searchEngvar, 'tobago') then
			useLang = 'TT'
	end
	
	return useLang and 'en-'..useLang
end

function addMaintCat(args)
	local catMaintenance
	if args.engvarCat ~= 'no' then
		local title = mw.title.getCurrentTitle()
		if title:inNamespaces(0) then  -- 0=main, 10=templ, 828=module
			if args.engvarCatSort then
				catMaintenance = '|' .. args.engvarCatSort .. ', ' .. title.text
			end
			catMaintenance = '[[Category:Articles using an unknown Template:Engvar option' .. (catMaintenance or '') .. ']]'
		end
	end
	return catMaintenance or ''
end

return p